Home / The Act / Chapter VI / Article 57

Article 57 — AI regulatory sandboxes

APPLIES AUG 2, 2027 WAS 2026-08-02 AMENDED BY OMNIBUS

Chapter VI · Governance
Source: EUR-Lex · CELEX 32024R1689
Verified 2026-07-16

PLAIN-LANGUAGE SUMMARY — ORIGINAL INTERPRETATION, NOT THE OFFICIAL TEXT

Each Member State must run at least one national AI regulatory sandbox — a supervised space to develop, train and test AI before launch. The Omnibus moved the deadline to 2 August 2027 (originally 2 August 2026). States may share a sandbox or join an existing one. Participants stay liable for damage they cause, but authorities impose no fines where a provider follows the agreed plan and official guidance in good faith.

AMENDED BY THE DIGITAL OMNIBUS ON AI

This provision's application date moved under the Omnibus. It now applies from 2027-08-02 (originally 2026-08-02). Source: Omnibus final text (OJ pending) · verified 2026-07-16

OFFICIAL TEXT Read verbatim on EUR-Lex ↗

The authoritative wording lives on EUR-Lex, quoted there in all 24 languages. We deep-link to the paragraph level rather than reproducing the full text. The consolidated post-Omnibus version will be linked the day its CELEX publishes.

AT A GLANCE

Chapter: VI — Measures in support of innovation

Applies: Aug 2, 2027

Tier: Governance

Guidance →Compliance timeline →Full Act text →

PENALTY EXPOSURE

—

of worldwide turnover — see Art. 99 penalties framework.

← PREVIOUS · ART. 56 Codes of practice NEXT · ART. 58 → Detailed arrangements for, and functioning of, AI regulatory sandboxes

ELSEWHERE IN CHAPTER VI — Measures in support of innovation

Art. 58 Detailed arrangements for, and functioning of, Art. 59 Further processing of personal data for develo Art. 60 Testing of high-risk AI systems in real world Art. 61 Informed consent to participate in testing in Art. 62 Measures for providers and deployers, in parti Art. 63 Derogations for specific operators